Retrieve a video generation

GET https://api.swishx.com/v1/videos/{id}

curl https://api.swishx.com/v1/videos/vid_8f2c1a9b3d4e \
  -H "Authorization: Bearer $SWISHX_API_KEY"

The video object

{
  "id": "vid_8f2c1a9b3d4e",
  "object": "video",
  "status": "succeeded",
  "model": "finsen-3.2",
  "prompt": "A pharma rep opens with a warm, confident smile.",
  "negative_prompt": null,
  "duration_seconds": 8,
  "resolution": "1080p",
  "routed_model": "veo-3",
  "keyframe": {
    "id": "img_3a9c1f2b8e7d",
    "status": "ready"
  },
  "compliance": {
    "verified": true,
    "flags": []
  },
  "output": {
    "url": "https://cdn.swishx.com/videos/vid_8f2c1a9b3d4e.mp4",
    "thumbnail_url": "https://cdn.swishx.com/videos/vid_8f2c1a9b3d4e.jpg"
  },
  "error": null,
  "cost_cents": 1200,
  "created_at": "2026-08-07T18:22:04Z",
  "completed_at": "2026-08-07T18:24:41Z"
}

Field	Type	Description
`id`	string	Unique identifier, prefixed `vid_`.
`object`	string	Always `"video"`.
`status`	string	`queued`, `processing`, `verifying`, `succeeded`, `failed`, or `canceled`.
`model`	string	The `finsen-3.x` version requested.
`prompt`	string	The prompt as submitted.
`negative_prompt`	string \| null	The negative prompt, if any.
`duration_seconds`	integer	Requested duration.
`resolution`	string	`"720p"` or `"1080p"`.
`routed_model`	string \| null	Which downstream model rendered this — see Model routing. `null` until routed.
`keyframe`	object \| null	The keyframe still used, if keyframe-first generation ran. `null` for direct text-to-video.
`compliance`	object \| null	`{ verified, flags }`. `null` until the output-verified stage runs.
`output`	object \| null	`{ url, thumbnail_url }`. `null` until `status` is `succeeded`.
`error`	object \| null	Populated when `status` is `failed`. Same shape as API errors.
`cost_cents`	integer \| null	What this generation cost, in cents. `null` until it resolves.
`created_at`	string	ISO 8601 timestamp.
`completed_at`	string \| null	ISO 8601 timestamp. `null` while still in progress.

Polling frequency

Poll every 2–5 seconds. Most generations resolve within a few minutes; there's no server-sent event or webhook yet, so polling is the only way to know when a generation finishes.

The video object

On this page